How do you track brand mentions in ChatGPT?

How do you do it by hand?

1. List the buyer questions: category (“best 2-person tent under $400”), comparison (“Halo 2 vs Tarn 2P”) and problem (“tent that holds up in rain”).
2. Ask each question in a fresh chat, with search on, signed out or with memory off, so earlier chats do not shape the answer.
3. Ask each one three to five times. Answers vary from run to run.
4. Record, for each answer: named or not, position, the words used about you, and every cited source.
5. Repeat weekly with the same questions, and compare rates, not single answers.

Where does the manual method break down?

It does not scale past a few dozen prompts, it is hard to keep runs consistent across people and weeks, and the sources are tedious to log. It also covers one engine at a time, while buyers use several.

What does a tracking tool add?

A tool runs the same prompts on every engine on a schedule, stores each answer with its sources, and turns them into rates you can trend. Lighthouse also shows which sources each engine cites for your category, which rivals are named in your place, and the page change that would most likely move the answer.

Where should you go next?

• What is AI visibility?
• How do AI engines choose their sources?
• Run a free AI visibility scan
• Prompt tracking in Lighthouse